Why do we see these changes in diagnostic criteria and concepts? Each iteration of the diagnostic manuals attempts to respond to the growing body of research, as well as trying to improve diagnostic processes. An example of the former is the decision to collapse the distinction between Asperger’s disorder and autistic disorder. A large body of work suggested few meaningful differences between these groups when current intellectual ability was equivalent (e.g. Eisenmajer et al., 1996; Macintosh & Dissanayake, 2004). Indeed, an influential study of diagnoses given across a range of expert clinics in America showed that the best predictor of which diagnosis was given (Asperger’s, Autism, PDD-NOS) was not any characteristic of the individual being diagnosed, but which clinic they went to! (Lord et al., 2012a). An example of a change in criteria reflecting ease of clinical use is the collapsing of social and communication criteria; it is hard to think of any piece of social behaviour a clinician might look for/ask about that does not involve communication, and vice versa. The overall aim of the changes in DSM-5 was to move to a system where a broad category diagnosis was accompanied by a detailed description of the individual’s strengths and needs, rather than trying to squeeze people into specific categories they didn’t necessarily fit. Do you get good customer responses when you're searching for SEO Services ?
Because the DSM tends to be more influential in research and therefore on psychological theory we will use this as the framework for subsequent discussions.
The variability in diagnostic manuals is nothing compared to the variability of presentation in the autistic population. When identifying autism, clinicians must be alert to the fact that the same feature may be manifest in dramatically different forms between individuals. For example, communication difficulties could mean that an individual is entirely non-speaking, speaking a great deal but mainly by echoing, or speaking fluently but with an atypical approach to conversational rules or understanding of non-literal language (e.g. irony, metaphor). Socially, a young autistic child may seem oblivious to others, while another individual on the spectrum may be keen to make friends but be unsure how to do so, making approaches that seem odd to neurotypical peers. Likewise, restricted and repetitive behaviours could mean lining up toys, spinning and flapping, a very ‘black and white’ thinking style, or might be evident as an immersive and impressively detailed interest in organic chemistry. One relevant dimension is clearly the presence or absence of intellectual disability (technically defined as an IQ less than 70 on a standardised assessment, accompanied by difficulties with daily living skills), but it is simplistic to suggest that IQ level alone determines how features are manifest.
As an example, let us consider the communication delays which are so prevalent in autism (Tager-Flusberg et al., 2005). A non-speaking autistic child may present with a significant intellectual disability which has contributed to their difficulties acquiring speech. Another non-speaking child may have no such intellectual barrier, but instead, their communication could be associated with anxiety-related mutism. Another factor which plays into this characterisation is the extent to which a feature presents obstacles in daily life. If either child can learn to communicate, for example via independent use of a text-to-speech device using visual symbols or Makaton signs, then this apparently profound difference might be minimally disabling (at least in environments where those communication modes are understood). Likewise, a specialist and all-consuming interest in geology could be something of a hindrance when trying to chat up a potential romantic partner (unless they are also a geologist!), but a boon when seeking employment in the mining industry.
Figure 3.1 The autistic constellation
In this three-dimensional space, we illustrate IQ scores, spoken language and a sensory feature often experienced by autistic people, as orthogonal dimensions. All data are hypothetical only. Autistic people may be located in every available point in the resulting three-dimensional space, but there is the potential, if we measure the right things, to identify clusters where features often overlap.
One attempt to visualise this complexity is shown above, where we try to illustrate autism as a constellation, rather than a spectrum.1 Here we show how a specific feature of autism (in this case, sensory hyper-sensitivity) might be plotted with intellectual ability and language profile. Autistic people, with a diagnosis, may locate themselves anywhere in the resulting three-dimensional space. Their exact location would further vary with context and across the lifespan. This space can be reproduced with different measures on each axis, using features which have relevance to theory, features more important for everyday life, or both in combination. For example, we might plot satisfaction with social relationships and against the number of social relationships and level of anxiety. Importantly, in this case, having few social relationships might cluster together with high relationship satisfaction and low anxiety. This group of people might be characterised as happy with a small number of friends, reflected in low anxiety scores. Another group might have few friends, but high anxiety and low satisfaction with their relationships. Could helping them achieve more social contact lower their anxiety? A third group might have high satisfaction, large amounts of social contact and high anxiety. Perhaps having a big social circle is both rewarding and stressful – could a better balance be achieved between the two?
1Thanks to Caroline Hearst who inspired our use of this term: www.autangel.org.uk/autism-constellation.html.
These examples illustrate that it is important to contextualise any such measurement within an individual’s own priorities – working with allies to identify these where the person themselves may find it hard to self-advocate. In the previous example, provided an individual is not at risk as a result of their relative isolation, a small amount of social contact should not be characterised as a sign of impairment, as in the first cluster described. The key take home message is that autism, both conceptually and in terms of lived experience, is manifest in complex interacting domains. To discuss the ‘autistic spectrum’ in a linear fashion, or to attempt to measure ‘severity’ vastly over-simplifies and misrepresents the reality. Where a measure of support need is relevant we propose using exactly that terminology – as it is in the DSM-5: e.g. Level 3: requiring very substantial support.
3. Making a diagnosis
Reliance on loosely defined diagnostic criteria presents challenges to differential diagnosis, risks of mis-diagnosis and difficulty matching a diagnosis with a relevant support package. Measures have been put in place to limit these problems. In the UK, the National Institute of Clinical Excellence (NICE) and the Scottish Intercollegiate Guidelines Network (SIGN) have both published criteria for a robust diagnostic process for autism. This should include multi- disciplinary assessment, direct observation of the individual across settings (e.g. clinic, home, school) and the use of standardised assessment tools. The widely used Autism Diagnostic Observation Schedule (now ADOS-2, Lord et al., 2012b) and Autism Diagnostic Interview (ADI-R, LeCouteur et al., 2003) are two such standardised diagnostic measures using direct observation and clinical history, respectively. Nevertheless, concerns remain about diagnostic practices. These tools are lengthy and costly to obtain and train in, making them impractical for low resource settings. Open-access diagnostic tools that are brief and can be administered by a range of people, are urgently needed, especially when one remembers that the vast majority of autistic people live in low- and middle-income countries. The utility of existing diagnostic tools for diagnosis in adulthood has also been questioned, as has the influence of such measures on the gender balance of diagnosed individuals (more on this later in the chapter). Despite flaws in the process, it should be emphasised that one function of the clinical diagnostic process is not only to offer an autism diagnosis when appropriate but also to take the opportunity to get to know the individual and their family. An in-depth diagnostic process has advantages, placing the clinician in a strong and informed position to make a diagnosis and, ideally, signpost relevant information and access to services for the future. Although we note that this is an ideal, rather than a reality, for many people, at the same time preliminary evidence indicates that having a diagnosis may yield benefits in terms of improved quality of life and reduced stress for the families of autistic people (McKechanie et al., 2017).
At present, a reliable, clinical diagnosis of autism is rare before the age of 3 years, though some argue it is possible as young as 18 months. This is primarily because the types of social behaviours that are characteristically different in autism (according to the diagnostic criteria noted earlier) do not emerge reliably in typical children until around three years. Moreover, restricted and repetitive behaviours are common in all children at about two years old (Leekam et al., 2007a). However, there is substantial interest in the possibility of pinpointing earlier indicators of autism. The search for very early signs that would allow one to predict which children would turn out to have autism, has been prompted by two different concerns. Practical considerations have pressed for earlier diagnosis in the hope that very early intervention might produce benefits on early stage outcomes like onset of language. Theoretical considerations urge the early identification of autism in order to explore the causal direction in development – for example, do differences in processing of faces lead to or result from difficulties in social interaction? The emergent findings on early signs will be reviewed in Chapters 4 and 7.
A relatively recent phenomenon is that people have begun to self-identify as autistic, often later in life and sometimes prompted by the diagnosis of a child in the family. There is little or no research evidence so far on the validity or consequences of this choice. However, it is certainly easy to understand why many people might recognise autism in themselves but either not feel the need for an external confirmation by a medical professional, or actively refuse to undergo a formal diagnostic evaluation that ends in being labelled with a ‘disorder’ (Kapp et al., 2013). At the same time, self-identification does raise concerns. Are there individuals who would not be independently identified as autistic, merely seeking to align themselves with a group that seems to be getting a certain amount of public attention? Are people who would benefit from mental health support mis-identifying the causes of their feelings? These cases are hopefully in a small minority. Nevertheless, the phenomenon of self-identification presents challenges to academic research, not least in efforts to estimate prevalence.
4. Prevalence estimates
How common is autism? Understanding this is crucial for service provision planning, including budgeting for the economic impact of autism, which is estimated to be very high in the UK (Knapp et al., 2009) and USA (Buescher et al., 2014; Lavelle et al., 2014). Kanner and Asperger both described the condition as rare. However, current prevalence estimates in Western countries tend to hover around 1% of the population. There is considerable global variation – a meta-analysis cited rates from 0.3% to 1.2% with a median rate of 0.6% (Elsabbagh et al., 2012) and called for more epidemiological work in low and middle-income countries. It is virtually impossible to trace the precise roots of this variability, but it is likely that differences in diagnostic procedures account for much of the range. In particular, we can see that those countries with less developed healthcare systems consistently produce the lowest prevalence estimates. Cultural differences likewise impact decisions post-diagnosis (Mandell & Novak, 2005) and should be taken into account within, as well as between, countries.
Accounting for the dramatic changes in autism prevalence estimates over time has been the focus of considerable media attention and academic effort (Fombonne, 2005). A range of headline-grabbing accounts have been put forward including the effects of pollution, changes to diet and, famously and tragically, vaccines. While it is known that there are environmental factors that contribute to autism – the condition is not 100% heritable – there is no robust evidence to support any of these accounts. In the case of the role of vaccines, this hypothetical causal factor has been thoroughly, rigorously and conclusively disproven (Jain et al., 2015; Taylor et al., 1999, 2014). Instead, variation in prevalence estimates of autism can probably be attributed to a combination of the following factors.
First, the diagnostic criteria for autism have changed dramatically since autism was first enshrined in diagnostic manuals (as ‘infantile autism’ in ICD-8, 1967; ‘childhood schizophrenia’ in DSM-I and II, and then ‘infantile autism in DSM-III in 1980). Specifically, they have broadened to admit a much wider variety of individuals under the diagnostic umbrella. Alongside this change, there has been a dramatic rise in awareness of autism, not just in the public but among medical professionals. If you visit your family doctor now to discuss concerns about your child’s development, autism will be a potential explanation on everyone’s radar (at least if your child is a boy!) in a way that was not the case 30 or even 20 years ago. This growing awareness, combined with broadened diagnostic criteria, may have given rise to a degree of diagnostic substitution as well. As autism diagnoses have increased, diagnoses of global developmental delay or intellectual impairment, have declined. This would imply that there is no absolute increase in the numbers of affected individuals, merely in the way they are being categorised. A fourth facet of this general process of raising awareness and widening categories, has been an increase in identification of autism in populations where that diagnosis was previously not considered. Diagnostic rates in adults are rising sharply, as adults with intellectual disability are re-assessed for autism as well, and as others start to recognise autism in themselves. Anecdotally, there seems to be a specific phenomenon of parents, and sometimes grandparents, seeking a professional opinion for themselves, following their child’s diagnosis.
Finally, in some cases differences in prevalence may result from differences in methodology. The Centers for Disease Control and Prevention in the USA recently published an estimate of 1 in 68, far exceeding any previous figure. However, the methods employed in this study have been criticised. There was no direct assessment involved – instead the estimate was based on prevalence of what appeared to be autism-linked features in the case notes of children and young people referred for evaluation by education or clinical service providers.
Is autism prevalence still increasing? There are a couple of UK studies that suggest that prevalence has reached something of a plateau over the last two decades, stabilising after the introduction of new diagnostic criteria in the early 1990s (Baxter et al., 2015; Taylor et al., 2013). Whether the recent changes in DSM-5 and ICD-11 will affect prevalence again remains to be seen. In addition, adult diagnosis and diagnosis among women and girls seem to be undergoing a recent sharp increase, which may not yet be represented in the latest epidemiological data.